| 基于贝叶斯神经网络的农田分区灌溉需水量模拟分析 |
| |
| 引用本文： | 朱爱华,戴光鑫.基于贝叶斯神经网络的农田分区灌溉需水量模拟分析[J].农业工程,2022,12(7):78-83. |
| |
| 作者姓名： | 朱爱华 戴光鑫 |
| |
| 作者单位： | 山东省海河淮河小清河流域水利管理服务中心，山东 济南 250100 |
| |
| 摘 要： | 针对农田分区灌溉需水量模拟过程中普遍存在的求解过程易陷入局部最小化、出现过度拟合，以及过度依赖历史用水数据，导致最终模拟结果存在显著误差的问题，研究基于贝叶斯神经网络的农田分区灌溉需水量模拟分析方法。以前一周需水量、年内月需水量占比、日内温度上限值及日降雨量为指标，通过聚类分析获取指标数据均值，对农田分区灌溉历史用水的样本数据进行聚类分析。构建贝叶斯神经网络模型，将指标数据均值输入模型，根据BP神经网络原理与贝叶斯规则训练指标数据，然后输出农田分区灌溉需水量模拟结果。试验结果显示数据聚类结果中数据间关联度高于95%，数据拟合效果较好，模拟需水量时具有更高的精度与稳定性。
|
| 关 键 词： | 贝叶斯 神经网络 农田分区灌溉 需水量模拟 数据聚类 |
| 收稿时间： | 2021/12/31 0:00:00 |
| 修稿时间： | 2022/2/24 0:00:00 |
Simulation analysis of farmland irrigation water demand based on Bayesian neural network |
| |
| Zhu Ai hu and Dai Guang xin.Simulation analysis of farmland irrigation water demand based on Bayesian neural network[J].Agricultural Engineering,2022,12(7):78-83. |
| |
| Authors: | Zhu Ai hu and Dai Guang xin |
| |
| Institution: | Haihe River，Huaihe River and Xiaoqinghe River Basin Water Conservancy Management and Service Center of Shandong Province，Jinan Shandong 250100，China |
| |
| Abstract: | In order to solve the problems of local minimization, over fitting and over dependence on historical water use data, which lead to significant errors in the final simulation results, a simulation analysis method of agricultural irrigation water demand based on Bayesian neural network is studied. Cluster analysis is carried out on the sample data of historical irrigation water use in farmland zoning. The water demand in the previous week, the proportion of monthly water demand in the year, the upper limit of daily temperature and daily rainfall are used as indicators, and the average value of index data is obtained through cluster analysis; The Bayesian neural network model is constructed, the mean value of the index data is used as the input of the model, the index data is trained according to the principle of BP neural network and Bayesian rules, and the simulation results of farmland zoning irrigation water demand are output. The experimental results show that the correlation degree between the data in the data clustering results of the research method is higher than 95%, the data fitting effect is better, and the simulation of water demand has higher accuracy and stability. |
| |
| Keywords: | Bayesian Neural network Farmland zoning irrigation Water demand simulation Sample data Data clustering |
|
| 点击此处可从《农业工程》浏览原始摘要信息 |
| 点击此处可从《农业工程》下载免费的PDF全文 |
